Presentation in Microsoft PowerPoint format

Traffic Hotspots in UMTS
Networks : influence on RRM
strategies
Ferran Adelantado i Freixer
([email protected])
COST 289
15-16 March 2004, Zurich
Outline
•Introduction
•Simulation environment
•Results
Path loss analysis
CAC performance
•Conclusions and future work
COST 289
15-16 March 2004, Zurich
Introduction
•The main goal of the study is to analyse non-uniformly traffic
distributed scenarios.
•It is important to be able to maintain the target QoS.
•All alternatives should be taken into account before deploying
hotspot WLAN networks.
•Assessment of RRM strategies becomes necessary to deal with
high traffic density areas (hotspots).
•Is it possible to dynamically react to environment changes?
COST 289
15-16 March 2004, Zurich
Simulation Environment
• A single isolated cell (radius R).
• A traffic hotspot with radius r and placed D meters from base
station.
• Ttotal=THS+TNo HS
where
THS=αTtotal
TNo HS=(1-α)Ttotal
•Only videophone users considered
•Propagation model:
Lp(d)=Lo+ log(d)
COST 289
a
R
D
15-16 March 2004, Zurich
Results
Simulation Parameters (1/2)
Cell radius
1000 m
Hotspot radius
50 m
UE parameters
Maximum transmitted power
21 dBm
Minimum transmitted power
-44 dBm
Mobile speed
10 km/h
BS parameters
COST 289
Cell type
Omnidirectional
Thermal Noise
-103 dBm
Pilot and common control
channel power
32 dBm
Shadowing deviation
3 dB
Shadowing decorrelation
length
20 m
15-16 March 2004, Zurich
Results
Simulation Parameters (2/2)
Traffic model
Call duration
120 seg
Offered bit rate
64 kb/seg (CBR)
Activity factor
1
Call rate
15 calls/h/user
QoS parameters
BLER target
1%
Eb/No target
2.95 dB
Propagation model
COST 289
Lo
128.1

37.6
15-16 March 2004, Zurich
Results
Impact of traffic distribution (1/5)
Path loss distribution
variation
Non-uniformly distributed
traffic scenario
BLER variation
Path loss pdf :
f Z ( z )  a  f ZHS ( z )  (1  a )  f ZNo HS ( z )
f ZNo HS (z ) :
no hotspot users
path loss pdf
f
hotspot users
path loss pdf
where
COST 289
HS
Z
(z )
:
15-16 March 2004, Zurich
Results
Impact of traffic distribution (2/5)
No hotspot users path loss :


A

e

2

R
f ZNo HS ( z )  
 Ae

2
 R
2
2
2
2
2
2
 1
 a  2  z 
e  1  erfc

2


 2
if z  a -  2 
1
 z  a  2 
e
erfc

2
2


if z  a -  2 
z
z
A  10
COST 289
2
Lo

 2
ln(10)

15-16 March 2004, Zurich
Results
Impact of traffic distribution (3/5)
Hotspot users path loss:
 Ae z
z

 1

HS
f Z (z)  
e 2  * 
2
 2 
  2r
2
2

 D 2  r 2  Ae z  
 
  2 arcsin
z


 
2

 2 D Ae
 
A  10
COST 289
2
Lo

 2
ln(10)

15-16 March 2004, Zurich
Results
Impact of traffic distribution (4/5)
Hotspot close to the base station
Hotspot far from the base station
Path loss pdf
Path loss pdf
Cell radius =1000m
D=150m
Cell radius =1000m
D=950m
0.035
0.035
0.03
a  0.0
0.03
a  0.0
0.025
0.02
a  0.3
0.025
a  0.3
a  0.7
0.02
a  0.7
0.015
0.01
a  1.0
0.015
a  1.0
0.01
0.005
0
0.005
0
75
95
115
135
75
85
95
Lp(dB)
105
115
125
135
145
Lp(dB)
Path loss pdf
Cell radius =1000m
a =0.2
0.025
D=
D=
D=
D=
D=
0.02
0.015
0.01
150m
350m
550m
750m
950m
0.005
0
75
85
95
105
115
125
135
145
Lp(dB)
Variation of hotspot location
COST 289
15-16 March 2004, Zurich
Results
Impact of traffic distribution (5/5)
a=0.0
a=0.3
a=0.5
BLER
1.53
1.86
2.04
HS BLER
N/A
2.63
2.60
No HS BLER
1.53
1.53
1.53
•As D increases, total BLER
increases.
•Hotspot users BLER grows for
large D.
•No hotspot users BLER is
lower for high D.
COST 289
•No hotspot users BLER is
maintained when increasing a
•Total BLER grows as a is
increased.
D=150m
D=550m
D=950m
BLER
1.46
1.48
2.04
HS BLER
1.00
1.07
2.60
No HS BLER
1.93
1.89
1.53
15-16 March 2004, Zurich
Results
Call Admission Control design (1/3)
Transmitted power for
mobile terminal
P
PT  L p N
1 
1
W 
 
R
1  b 
 Eb 


 N 0 T
Outage probability in UL

 W 


  

  Eb   Eb  
PT max 1      Rb 


  
   p LP 
p 

1
 E 

PN
b
  N 0   N 0 T 


 
 

N
  0 T


max  1 
Maximum
admission threshold
for a certain Lp
COST 289
L p*PN
1
PT max  W 


R 
 b  1
 Eb 


N 
 0 T
15-16 March 2004, Zurich
Results
Call Admission Control design (2/3)
Admission threshold may be determined with Path Loss statistics (Cumulative
density function) :
a
max
0.0
0.77
0.5
0.68
Outage probability = 0.5 %
BLER ≈ 1.3 %
BLER can be maintained by adjusting max
BLER Hotspot(%)
BLER (%)
2.2
1.7
a0
1.6
0.77
a0
2
a0.5 0.68
a0.5 0.77
1.5
0.77
a0.5 0.68
a0.5 0.77
1.8
1.4
1.6
1.3
1.4
1.2
1.2
1.1
1
1
30
35
40
45
50
55
Number of users
COST 289
60
65
70
30
35
40
45
50
55
60
65
70
Number of users
15-16 March 2004, Zurich
Results
Call Admission Control design (3/3)
Maintaining low BLER with hotspots leads to an admission
probability decrease.
Admission probability
105
100
95
90
85
80
75
70
65
60
a0 0.77
a0.5 0.68
a0.5 0.77
30
40
50
60
70
80
Number of users
COST 289
15-16 March 2004, Zurich
Conclusions and Future Work
•In non-uniformly distributed traffic scenarios, without applying CAC,
hotspots with high D and a cause a QoS degradation.
•Suitable admission control threshold (max) can be determined if path loss
statistics are known.
•Maintaining low BLER implies an admission probability decrease.
•Future work will be focused on dynamic hotspot detection.
•Design and assessment of adapted RRM strategies will determine if it is
necessary to include a hotspot WLAN .
COST 289
15-16 March 2004, Zurich